Tidy summarizes information about the components of a model. A model component might be a single term in a regression, a single hypothesis, a cluster, or a class. Exactly what tidy considers to be a model component varies cross models but is usually self-evident. If a model has several distinct types of components, you will need to specify which components to return.
# S3 method for roc tidy(x, ...)
Additional arguments. Not used. Needed to match generic
signature only. Cautionary note: Misspelled arguments will be
tibble::tibble() with columns:
The cutoff used for classification. Observations with predicted probabilities above this value were assigned class 1, and observations with predicted probabilities below this value were assigned class 0.
The false positive rate at the given cutoff.
The true positive rate at the given cutoff.
data(churn)#> Warning: data set ‘churn’ not foundr <- roc(churn$predictions,churn$labels)#> Error in roc(churn$predictions, churn$labels): could not find function "roc"td <- tidy(r)#> Error in tidy(r): object 'r' not foundtd#> Error in eval(expr, envir, enclos): object 'td' not found#> Error in ggplot(td, aes(fpr, tpr)): object 'td' not found# compare the ROC curves for two prediction algorithms library(dplyr) library(tidyr) rocs <- churn %>% gather(algorithm, value, -labels) %>% nest(-algorithm) %>% mutate(tidy_roc = purrr::map(data, ~tidy(roc(.x$value, .x$labels)))) %>% unnest(tidy_roc)#> Error in eval(lhs, parent, parent): object 'churn' not found#> Error in ggplot(rocs, aes(fpr, tpr, color = algorithm)): object 'rocs' not found